🚀 go-pugleaf

Thanks Richard! You have been great help..
I'm just abit concerned about the corner cases, but it should be ok! :)

Richard Heathfield <binary@eton.powernet.co.uk> wrote in message news:<3BB2C06D.5537E07E@eton.powernet.co.uk>...
> Adam wrote:
> > 
> > When you say "portable" most systems have wininet.dll installed.
> 
> I just checked my Linux installation and hey, what do you know, it's not
> there. In comp.programming (which is where I'm reading this), "portable"
> means more than just "portable to Windows".
> 
> > It is part
> > of the OS in Win '98 and >. But for users with Win '95 - all they need is IE
> > 4 or >.
> 
> <shrug>
> 
> > "justin chan" <jayc@start.com.au> wrote in message
> > news:16379a2c.0109251629.5566caa3@posting.google.com...
> > > Hi,
> > >
> > > Currently I'm using 'InternetCombineURL' function from WinInet to
> > > combine a base and relative URL into a single URL.
> > > eg. Base: http://www.myserver.com/folder/
> > >     Relative: ../file.html
> > >     Result: http://www.myserver.com/file.html
> > >
> > > However, because I want to make my application portable, I'm removing
> > > dependencies on WinInet.
> 
> Excellent.
> 
> > >
> > > Does anyone know if there is an alternative to 'InternetCombineURL'
> > > and does not rely on WinInet??
> 
> Yes, Justin, it's called "programming".  :-)
> 
> Start off with your base URL. Parse on the / character. This gives you
> tokens http: (in your example), www.myserver.com, and folder.
> 
> Build a data structure that represents that relationship:
> 
>       www.myserver.com
>             ^
>             |
>             |
>             |
>             V
>           folder
> 
> This rather simple example doesn't explain things very well, but
> basically you'd have each node pointing to an array of child nodes and
> also to a parent node (with the server node itself having a parent of
> NULL). That would enable you to build up a picture of the server as you
> go.
> 
> Anyway, from that start, finding a relative path is trivial. Start at
> the "current" directory - in your case, www.myserver.com/folder - and
> parse, again on the / character. If you hit a . on its own, you can
> discard it. If you hit on .. on their own, you need to go one place
> nearer the root. If you hit a token terminated by a /, you need to
> change into that directory (possibly creating a new node in your data
> structure in order to represent it), and if you hit a non-empty token
> not ending in / at the end, that's the file you're after.
> 
> It's really relatively simple, but also relatively interesting, and well
> worth the exercise of programming for yourself.